The present invention relates to network node and network connection, and more specifically, to a method and apparatus of managing network connection of a network node.
A traditional computer network is consisted of a plurality of interconnected computers, in which each physical computer entity acts as a node of the network. With the development in virtualization, a physical computing platform may have hypervisor running thereon and then have Virtual Machines installed on the hypervisor. A Virtual Machine (VM) is capable of providing various applications and services just like a real standalone computer. Since a VM acquires its running environment completely via software simulation, a plurality of VMs may be installed on a same physical computing platform, and the plurality of VMs may run concurrently independent to and isolated from each other. In this case, such physical computing platform is often regarded as a network node. In cloud environment, a network is typically formed by interconnecting a plurality of physical computing platforms as above.
FIG. 1 illustratively shows two network nodes in a network. The two network nodes are shown as node 1 and node 2, respectively. As mentioned above, several VMs may be installed on each node to provide necessary applications and services. As is known to those skilled in the art, a node connects to the network via a network card, that is, network interface card (NIC). In a typical network environment, as shown in FIG. 1, each network node at least has two types of NIC, that is, NIC 0 and NIC 1, wherein NIC 0 is for an administrator to conduct network management and is also referred to as administration NIC; NIC 1 is for transmission of traffic data of VMs and is also referred to as traffic NIC. Both administration NICs of the two nodes are connected to an administration network to transmit internal administration data including, for example, administration configuration data, API invoke instructions, image data, etc., while both traffic NICs are connected to a VM traffic network to transmit business traffic data of the VMs. Inevitably, after being used for a period of time, a NIC has certain failure probability. For example, assuming NIC 1 for traffic data on a node fails, all VMs carried by that node will be inaccessible from outside at this point, and data transmission related to applications running on these VMs will be interrupted, thereby influencing provision of VM services.
To avoid or alleviate influence caused by failure of a NIC, in prior art, data of the failed NIC is transferred by methods such as multi-NIC in group, NIC bonding, link aggregation, etc. However, these methods basically require additional NIC and/or switch in hardware. In practice, it is quite difficult to complete hardware switch or replacement in a short time. Furthermore, it takes a certain time window for an administrator to become aware of the NIC failure. Therefore, using these methods in the art, it is difficult to solve the problem that data of the failed NIC would be interrupted before the administrator comes to know the failure of the NIC and to complete the hardware replacement.